[2.5] Registering a KVM host provides insufficient debugging information on failure
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
Invalid
|
High
|
Unassigned |
Bug Description
After attempting to deploy a KVM host with MAAS, I was presented with a "Failed deployment" message. In the regiond.log, I found the following exception:
https:/
This does not provide enough information to diagnose the problem. We should consider the following:
(1) Is it possible to add additional retries to the KVM host deployment process?
(2) Can we add additional logging in failure cases, such as which IP address was chosen as the pod URL, and the IP addresses configured on the machine?
(3) Should we try additional URLs in case of firewall issues?
(4) Should we consider more opinionated ways to designate network spaces? For example, if we know that a particular space is for network management rather than data plane traffic, that's likely to be a better space to select for network management purposes.
I didn't expect any retries to be necessary here, since we already have a "sleep 10" at the end of the array of "runcmd" options in the cloud-init vendor-data we use for deploying KVM hosts. It may be possible that a different race condition is interfering with the completed deployment, though.
description: | updated |
Changed in maas: | |
milestone: | 2.5.0rc1 → 2.5.0rc2 |
Changed in maas: | |
milestone: | 2.5.0rc2 → 2.6.0 |
I just installed MaaS 2.5 beta 4, I can deploy Ubuntu 18.04 Machines normally, but, when I click to deploy the machine "As a KVM Host", it fails and I can't easily find why.